How to determine input and target data for classification in neural network

Hi all,

I want to classify using neural network toolbox. I have input matrices data (7 input x 125 samples) and target data (5 output x 125 samples). I have scaled my input data that have 125 samples into 5 class. Row 1:25 as class A, row 26:50 as class B, row 51:75 as class C, row 76:100 as class D, and row 101:125 as class E. I want to use 'trainlm' training. I am confuse to write input and output in Matlab.

Thanks in advance for any and all help

Answer 1

[Greg Heath]

%size(input) = [ 7 125 ]

%size(target) = [ 5 125 ]

 target = zeros(5,125);

 target(1,1:25) = ones(1,25);

...

target(5, 101:125) = ones(1,25);

Use patternet.

Hope this helps

Thank you for formally accepting my answer

Greg

[Looi]

Mind to know can I use back propagation instead of patternet?

[Greg Heath]

What do you mean by backpropagation ... fitnet?

Fitnet is for regression and curve-fitting. Targets are real-valued and trainlm is used.

Patternnet is for classification and pattern-recognition. Targets are columns of eye(c) where c is the number of classes or categories. Trainscg is used because targets are binary.

Except for that and plotting, the algorithms are equivalent (both call feedforwardnet).

Why do you have a problem with that?